Leveraging Probabilistic Modeling for Robust End-to-End Autonomous Driving across Domains

End-to-end (E2E) autonomous driving has recently emerged as a new paradigm, offering significant potential. However, few studies have looked into the practical challenge of deployment across domains. In this work, we propose \ours, a novel framework for \underline{Ro}bust \underline{C}ross-domain E2E \underline{A}utonomous driving. \ours formulates the joint probabilistic distribution over the tokens that encode ego and surrounding vehicle information in the E2E pipeline. Instantiating with a Gaussian process (GP), \ours learns a set of basis tokens with corresponding trajectories, which span diverse driving scenarios. Then, given any driving scene, it is able to probabilistically infer the future trajectory. By using \ours together with a base E2E model in source-domain training, we improve the generalizability of the base model, without requiring extra inference computation. In addition, \ours enables robust adaptation on new target domains, significantly outperforming direct finetuning. We extensively evaluate \ours on various cross-domain scenarios and show that it achieves strong domain generalization and adaptation performance.